Feature Requests

The ability to upload 100 (or even 1000...or more) of my own letters so that it could learn to mimic how I write

If as a clinician I asked - are you still continuing to have pain ... the AI generated summary mistook this as patient saying they are continuing to have pain.

It would be helpful if the context could remain the same within a specific patient's chart, rather than having to add context every session.

Until EMR integration is in place Instead of tediously entering one patient at a time and cancelling recording Need to be able to enter list of patients per date/time

I see the Memory tab of Heidi but unless I am misunderstanding how it's used, it won't quite work as I'm requesting here. I have consistent requests for how I would like individuals to be referred to, how I hope for the note to be formatted, etc., and despite having these instructions explicitly stated in my Heidi note template, it will routinely go against those instructions at random times, then I need to prompt Heidi to rewrite it. For example, I do not want names used and have tried various ways to indicate that in the template, but at times it will still use them. For now, I have just gone about regularly "correcting" Heidi, rating the generations, etc. but it would be wonderful if we could give general rules to follow. It may be a long shot as even with the advanced ChatGPT and its newer "memory" function, it still fails on very basic instructions, but in case Heidi AI is different than ChatGPT in ways that would make it work, I wanted to submit this request. Thanks!

Dividing the outputs into distinct sections allowing me to copy, dictate into, and regenerate using Ask Heidi AI only on the specific parts.

I noticed in the last week or so Heidi seems to have lost the ability to have leading spaces at the start of a line in order to show that something is a list. So whereas before it would produce stuff like this: Heading 1 - item 1 - item 2 Heading 2 - item 3 - item 4 Now it looks like this: 1 Heading item 1 item 2 Heading 2 item 3 item 4 It seems like the text is still generated the same so I'm wondering if this is something happening AFTER the LLM is generating the text, that the leading spaces are getting deleted somehow when Heidi displays the output from the LLM? It definitely makes the notes harder to read and also triggers my OCD so I end up wasting time putting the spaces in. Edit: looks like the lists get automatically reformatted here so I'll attach a screenshot to show what I mean.

It is tiresome to go through notes and update every contraction, and as tiresome to add them all as find and replace. As for He's, this could be he is, or he has so a find and replace would not be meaningful.

Allot of important information can get left out if it is not specifically spoken about early in the visit.

Uploading a series of checklist questions which will be detected based on the Transcription and chief complaint: perhaps harnessing the memory function and using if-then statements to ensure important questions are not missed. Example: let’s say, I saw a patient for weakness and asked all the standard questions except those that would be important in someone with a myopathy. Heidi would detect the absence of these questions based on the transcription and comparing it to the previously uploaded checklist for “muscle weakness“ and suggest to the provider additional questions to be clarified on the history to narrow the differential.